Recovery from spirals with the LVL button

In a flight in a Cirrus SR22, it was mentioned in passing that the LVL function on the Garmin autopilot is not taught for unusual attitude recovery. A flight in the RV-9A, equipped with a Garmin G3X Touch system, was then made to evaluate the LVL function for spiral recovery. After data from the first flight was analyzed, it was determined to be safe to fly more aggressive spirals on the second flight.

Two key observations were made from the RV-9A flights:

  1. The LVL function applies roll and pitch inputs simultaneously. The FAA book technique (next section) is to level the wings first and to then recover in pitch;
  2. The Garmin LVL function apparently has two modes, one more aggressive than the other. The key parameter is time to wings level.

Additional test points were flown by another pilot in a Vashon Ranger LSA with Dynon avionics. The Dynon recoveries also seemed to apply pitch and roll corrections simultaneously. Recoveries on average took twice as long, and there were two recoveries with oscillatory components. Those are discussed below.

An additional flight was made to compare book recovery technique from spiral (roll level, then pitch up) with the simultaneous pitch/roll used by the autopilots. Significant differences were observed in altitude loss and speed gain. No roll tightening was observed with the simultaneous pitch/roll technique.

While these flight tests do not indicate that the FAA-recommended spiral recovery is flat out wrong, these flight tests clearly indicate that the FAA technique is not always required for all airplanes. These flight tests also suggest that the FAA technique be re-examined for a large number of airplanes and that airplane-specific characteristics be noted, just as wing drop in stall recovery techniques vary from airplane to airplane. (A few planes will spin with aileron input in the stall, some aerobatic planes will do a nice snap roll with rudder input in a stall, many modern airplanes have good aileron authority in a stall with little danger of spin.)

Background: Quotes From FAA Publications

(c) Graveyard spiral. An observed loss of altitude during a coordinated constant-rate turn that has ceased stimulating the motion sensing system can create the illusion of being in a descent with the wings level. The disoriented pilot will pull back on the controls, tightening the spiral and increasing the loss of altitude.

It is important to note that some training airplanes will not enter into the developed phase [of a spin] but could transition unexpectedly from the incipient phase into a spiral dive. In a spiral dive the airplane will not be in equilibrium but instead will be accelerating and G load can rapidly increase as a result. 

Spiral Dive

A spiral dive, a nose low upset, is a descending turn during which airspeed and G-load can increase rapidly and often results from a botched turn. In a spiral dive, the airplane is flying very tight circles, in a nearly vertical attitude and will be accelerating because it is no longer stalled.

  1. Reduce Power (Throttle) to Idle
  2. Apply Some Forward Elevator
  3. Roll Wings Level. Roll to wings level using coordinated aileron and rudder inputs. Even though the airplane is in a nose-low attitude, continue the roll until the wings are completely level again before performing step four.
  4. Gently Raise the Nose to Level Flight
  5. Increase Power to Climb Power

RV-9A Flight Test Data with Spiral Recovery Using the G3X Touch LVL Function

A video of the first flight tests is available here. Autopilot engagement is indicated by “LVL LVL” appearing above the roll pointer near the top of the display.

First flight, with spirals in alternating directions:

Second flight, all spirals to the left:

Analysis of the Garmin G3X Touch LVL Function in RV-9A Flight Tests

  1. These tests indicate satisfactory use of the G3X Touch LVL function for the points tested;
  2. None of these tests used max available programmed roll torque (40%) or pitch torque (80%);
  3. The Max G numbers are not necessarily precise because of sensor noise. However, the data suggest that the LVL function is limited to 1.8 Gs. The cockpit G meter, more heavily damped, never exceeded 1.8 Gs;
  4. AOA was never a factor. Highest AOA seen on the recorded data was 0.3, and stall is 1.0.
  5. The pitch and roll servo torques peaked at about the same time;
  6. As discussed above and as seen in the video, the LVL function apparently has two modes, a lazy mode for minor upsets and a friskier mode for larger upsets. Time to level flight is in the tabular data in the last column;
  7. Data sampling was only available at 1 Hz, and this may be responsible for data artifacts.

Vashon Ranger Autopilot Recovery Flight Test Analysis

A similar series of flights were conducted in a Vashon Ranger LSA equipped with Dynon avionics and autopilot. Spiral recoveries were flown from 30-degree, 45-degree and 60-degree banks, both level and nominally 10-degrees nose down, at 60 KIAS and 90 KIAS. Each entry was flown in both left and right spirals.

The recorded data did not include autopilot engagement, and roll and pitch force data were so quantized as to convey little information. The aircraft roll and pitch attitude information was recorded at 4 Hz. However, roll and pitch data, combined with the experience of the Garmin flight data, allowed reasonable conclusions to be drawn.

  1. The Dynon spiral recovery applied roll and pitch inputs simultaneously, like the Garmin;
  2. The Dynon recoveries took about 10 seconds, about twice as long as most Garmin recoveries;
  3. The Dynon autopilot, as installed in and configured for this particular airplane, did not always perform optimally. Most recoveries were nominal, but at two data points, pitch and roll had damped oscillatory components.

RV-9A autopilot settings as flown:

Spiral Dynamics and Manual Recoveries

On a previous flight, it was determined that in a stick free spiral, the nose would come back up to the horizon, presumably due to speed stability. That spiral was terminated when the nose came back up for cautionary reasons. A flight was made to explore this further with a manually flown spiral, starting at 60 degrees of bank and 10 degrees nose down. That must be re-flown, and flights with more nose down may also be flown.

Two steep spirals were flown to compare book recovery technique (roll, then pitch) with the simultaneous roll and pitch recovery technique. The tests started at 71 KIAS or so in level flight. Nominal spiral entry points were 60 degrees of bank and 30 degrees nose down. On the book recovery test, the pitch bobbled slightly lower as roll was entered, probably due to pilot error. Statistics for the two spirals are:

The book recovery technique kept the nose down longer, resulting in a significantly greater airspeed during the recovery and a spectacularly greater altitude loss. Note the g achieved during recovery, more than was anticipated.

The significant parameters of these two tests more clearly show the differences in pitch during the recovery, leading to the above-mentioned airspeed and altitude differences.

Because of the 2.3 and 2.45 g pulled in these recoveries, and because previous flight tests suggest that the G3X Touch autopilot may have a 1.8 g limit, there are currently no plans to explore the LVL function in any spirals steeper than those already flown. Also, it is not clear what benefit would be provided by fully documenting the performance of this one system in this one airplane with this one configuration of torque and gain settings.


    • And almost all autopilots have those two buttons already…I have practiced this. I am willing to bet most pilots have not…!

  • The point of the exercise was to explore the LVL function as a one-stop remedy to unusual attitudes, in this case, spirals.
    * The LVL function algorithms appeared to be different from the regular autopilot algorithms;
    * In this system, pushing HDG will command the autopilot to turn the airplane to whatever heading is set on the heading bug. This is not appropriate for a spiral recovery.

  • a LVL button should not be used an unusual attitude recovery crutch … basic airmanship, not an autopilot, should be used.

    • Agree with you on a theoretical level, Matt, but emergency equipment is there to provide a life-saving out when “basic airmanship,” for whatever reason, isn’t enough.

      • Over reliance on automation has led to complacency, loss of aircraft and loss of life. Teaching someone to push a button to get out of an unusual attitude might be great for a “pinch hitter” course but is absolutely the wrong course of action for a licensed pilot. If anyone’s personal course of action is to push a button to recover from an unusual attitude, they shouldn’t be a licensed pilot and most likely didn’t have the passion to pursue becoming an aviator in the first place. Someone that chooses this course of action should choose intelligence over pride and hire a pilot / CFI to fly for them or with them always.

  • Ed: Please be careful pitching and rolling at the same time. The Garmin/Dynon equipment are probably designed with structural margin in mind – hence the 1.8G limit (probably 2G limit). Certificated airplanes are not designed to pull maximum G load and roll at the same time. Va will protect a pilot from a single-axis, full throw control input; not a multiple axis full throw maneuver. This is similar to the Airbus loss of the vertical stabilizer off the east coast. Full rudder input is covered, a doublet was not.

    I enjoy your work in this area. Knowing this data could result in less LOC in many base-to-final scenarios.

    • Agreed. The rule of thumb is that for a rolling pull up, reduce the designed G load by a third. Using a 3.8 g design limit load for the RV-9A, that gives 2.5 g limit load for a rolling pull up, without going into the ultimate load safety cushion. I don’t know what the RV-9A was really designed for, but I covered my bases with successive flights, analyzing the data for one flight before I went on to the next flight. And I did not explore potential tightening up of the spiral with back stick because I felt there was risk of getting more g than anticipated and a rolling pull up recovery.

      And thanks for your kind words.

  • Ed: Well, you don’t comment on application of g force while rolling. As most tactical jet pilots will tell you, rolling g’s stress the aircraft considerably more than g’s that are orthogonal to the lateral axis. Yeah, tactical jets pull more g’s than GA aircraft, but when someone is pulling out of an unusual attitude, arm strength is dependent on Adrenalin.

    I don’t doubt that the automatic techniques result in less loss of altitude. But the point is that the FAA technique (i.e., roll first and then pull) is actually simpler and will put less stress on the aircraft.


    Vince Massimini Kentmorr Airpark, MD

    • Please see my previous comment.
      The point of this investigation was to see what a LVL function actually did and to see what could be learned from it.
      And while the FAA technique can, in the one sense reduce combined g force due to pitch and roll, the FAA technique can allow airspeed to increase with the possibility of increased g from the higher airspeed — especially if an overwhelmed pilot delays recovery initiation.
      Other research, not mentioned here, documents that low speed spirals can masquerade as spins. A base to final low speed spiral with a slow recovery technique may run out of altitude.
      Bottom line is that spirals are an involved topic and are worthy of more attention than the simple answers currently accepted.

  • I’m curious if the plane is inverted in IMC and the disoriented pilot chooses to engage LVL, will the plane right itself?

    • I’ll let somebody else explore that situation… If they’re inverted, they’re probably also nose down, fast and accelerating.

      • I have a GFC500 in a Cessna and I’ve never chosen to be inverted. With the EPS active it’s doubtful that could happen accidentally.

Leave a Reply

Your email address will not be published. Required fields are marked *